Emission factor of ammonia (NH₃) from on-road vehicles in China: tunnel tests in urban Guangzhou

NH₃ emission from motor vehicles can be influenced by many factors, such as vehicle and fuel types, emission control technology and driving patterns (Moeckli et al 1996, Fraser and Cass 1998, Kean et al 2000, 2009, Baum et al 2001, Durbin et al 2002, Heeb et al 2006, 2008, Livingston et al 2009). NH₃ emission factors for vehicles can be estimated in road tunnels (Moeckli et al 1996, Fraser and Cass 1998, Emmenegger et al 2004, Kean et al 2000, 2009) or obtained from classis dynamometer studies (Durbin et al 2002, Huai et al 2003). Road tunnels are excellent locations to measure vehicle emissions as the influence of meteorological conditions is minimized and the emissions undergo 'real world' dilution effects with minimal interference from other sources, enabling the vehicle emission factors to be determined. More importantly, road tunnel studies provide emissions from a large number of vehicles under 'real world conditions', that is, vehicles operating at hot and stable conditions when going through a road tunnel.

We conducted our tests in the Zhujiang Tunnel. It is an underwater tunnel crossing the Pearl River in the western urban area of Guangzhou, the central city in the Pearl River Delta. Figure 1 shows the layout of the tunnel. The tunnel is 1238 m long with a 721 m flat underwater section and two 517 m open slope sections outside both ends; it has two traffic bores of two lanes with traffic in the same direction each (He et al 2008). The vehicle speed limits in the tunnel are 15–50 km h⁻¹, and the typical vehicle speed is 40–50 km h⁻¹. It is worth noting that during rush hours (7–9 am and 5–8 pm) all trucks are banned; trucks of Guangzhou with vehicle payload weight higher than five tons and trucks from other cities with vehicle payload weight higher than 0.6 tons are banned from 7 am to 10 pm. During this study, the ventilation fans in the tunnel were turned off to prevent the influence of air turbulence.

Zoom In Zoom Out Reset image size

A video camera was placed at the exit to record the passing vehicles. Numbers of different types of vehicles were determined from the video tapes. Motor vehicles are classified into taxi, cars, mini-trucks (MTs), light-duty trucks (LDTs), medium-duty trucks (MDTs), heavy-duty trucks (HDTs), light-duty buses (LDBs), medium-duty buses (MDBs), heavy-duty buses (HDBs) and motorcycles. During the sampling period, the traffic densities varied between 105 and 2641 vehicles per hour, with an average of 1782 vehicles per hour.

Sampling points were located 50 m from both ends of the tunnel inside the western bore with a cross-sectional area of 52.8 m². Sampling time lasted from 3 am 10 August to 0 am 15 August 2013, with an average temperature of 33.4 ± 3.4 °C and relative humidity of 53.9%. Measured pollutants included NH₃, NO_x , CO and CO₂.

NH₃ and NO_x were continuously measured by a chemiluminescence analyzer (EC9842, Ecotech, Austria). The EC9842 chemiluminescence NH₃ analyzer combines a high performance ammonia converter (HTO-1000) and proven chemiluminescence detection to measure NH₃, NO_x and total nitrogen compounds (N_x ). N_x is the sum of NO, NO₂ and NH₃. To measure N_x concentration, NO₂ and NH₃ are converted to NO in a quartz converter by Pt catalyst at a temperature range of 580–820 °C. In a separate reaction, NO_x is passed through a molybdenum converter heated to approximately 325 °C to convert NO₂ to NO. NO is then detected by the analyzer. The resulting NH₃ concentration is determined by subtracting the N_x result from the NO_x . Measured NH₃ concentration may be interfered by basic gases, such as organic amines or the like. McCulloch and Shendrikar (2000) compared experimental results of concurrent sampling using this method with a Model 17C analyzer (Thermo Environmental Instruments, Franklin, MA) and collocated citric acid denuders (URG, Chapel Hill, NC). Their results demonstrated that this method reliably measured ambient NH₃ concentrations with a high degree of accuracy. This method was also successfully employed to measure NH₃ in the southeastern US (Hansen et al 2003, Edgerton et al 2007). The instrument was calibrated using a certified cylinder of NO (Foshan Kodi Gas Chemical Industry, China). NH₃ converter efficiency was determined to be 91.28% using a certified cylinder of NH₃ (Beijing AP BAIF Gases Industry, China). The detection limit and accuracy of the NH₃ instrument are 0.5 ppb and ±0.5%, respectively, and the time resolution is 1 min.

For accurately determining CO₂ and CO concentrations during a time interval, two-hour integrated air samples were collected by pumping air into the cleaned aluminum foil bags at a constant rate of 100 ml min⁻¹. Samples were collected every two hours from 6 am 10 August to 4 am 11 August and from 4 am 12 August to 0 am 13 August for two days. CO concentrations in these air samples were determined using a gas chromatography (Agilent 6980GC, USA) with a flame ionization detector and a packed column (5A Molecular Sieve 60/80 mesh, 3 m × 1/8 inch). Details of this instrument can be found elsewhere (Zhang et al 2012b). CO₂ was analyzed with HP 4890D gas chromatography (Yi et al 2007). For CO and CO₂ the detection limits were <30 ppb and the relative standard deviations were all less than 3% based on seven duplicates running with CO and CO₂ standards of 1.0 ppm.

Concentrations of NH₃ and NO_x measured at the tunnel inlet and outlet are presented in table 1. Averaged concentrations of NH₃ and NO_x measured at the outlet were 13.8 and 8.0 times that at the inlet, respectively. The average concentration of NH₃ at the tunnel outlet was 729 ppb, much higher when compared to ∼390 ppb reported by Moeckli et al (1996) at the outlet of Gubrist Tunnel for a one-day exposure period, or 34 μg m⁻³ (48.9 ppb NTP) measured by Fraser and Cass (1998) inside the Van Nuys Tunnel, or 369 ppb observed by Kean et al (2000) at the exit of Caldecott Tunnel from 4–6 pm.

As shown in figure 2, NH₃ and NO_x at the tunnel outlet all exhibited diurnal patterns quite similar to that of the total traffic density. Each day both NH₃ and NO_x concentrations had two peaks at the rush hours of 7–9 am and 5–8 pm, and they were higher during evening rush hours than during the morning ones, consistent with about 10% higher total traffic densities during evening rush hours. The maximum concentrations occurred during the evening rush hours with levels over 4000 ppb and 2500 ppb for NO_x and NH₃, respectively. The concentration valleys were reasonable during the nighttime of 0–6 am when traffic density was also the lowest.

Zoom In Zoom Out Reset image size

Figures 2(b) and (c) show the traffic density profiles for taxi, MTs, LDTs, MDTs, HDTs, cars, LDBs, MDBs, HDBs and motorcycles inside the tunnel. Diurnal patterns of cars and LDBs exhibited two peaks in rush hours, the same as that of the total traffic density. Taxi showed a quite different pattern with a peak at midnight. Densities of MDTs and HDTs showed no significant differences between rush hours and other time periods. As MDTs and HDTs only contributed about 0.5% of the total traffic, the restrictions on trucks might have a negligible effect on the composition of vehicle fleet. Percentages of light-duty (LD) vehicles (cars, taxi, MTs, LDTs and LDBs) were relatively stable during 10 am–10 pm (figure 2(d)) and the sum of cars, LDBs and taxi contributed approximately 82.5% of the total traffic, which was quite near a share of 79.6% for small passenger vehicles in all civil vehicles in Guangzhou in 2012 (Guangdong Provincial Bureau of Statistics (GPBS) and Guangdong Survey Office of National Bureau of Statistics (GSONBS) 2013). This consistency means, if cars and LDBs were major contributors of NH₃ as discussed below, our tunnel test results would be largely representative ones for Guangzhou.

Figure 3 shows the scatter plots of NH₃ concentrations against traffic densities of different types of vehicles. NH₃ concentrations show a good positive correlation with the total traffic density. Poor correlations between NH₃ concentrations and densities of diesel trucks were probably related to the negligible ammonia emission from diesel vehicles (Burgard et al 2006). Although taxi contributed about 14.3% of the total traffic in the tunnel from 10 am to 10 pm, its density was also poorly correlated with NH₃ concentration. A probable reason is that NH₃ concentrations varied with that of major NH₃ contributors like cars, and that ammonia emission from taxi was negligible as over 90% of taxis in Guangzhou were LPG-powered instead of gasoline-powered. The negative correlation between motorcycle densities and NH₃ concentrations might be resulted from the fact that motorcycle densities peaked during nighttime when densities of major NH₃ contributors, like cars, had valleys instead. The good positive correlations (p < 0.001, R² > 0.5) between NH₃ concentrations and densities of cars and buses (LDBs, MDBs and HDBs) might imply that cars and buses were largely responsible for NH₃ emissions in the tunnel.

Zoom In Zoom Out Reset image size

Light-duty vehicle emission factors were calculated by carbon balance, using the following equation the same as previous studies (Kirchstetter et al 1999, Kean et al 2000, Martins et al 2006):

$$\begin{eqnarray}&&{{E}_{P}}=\left( \frac{\Delta [P]}{\Delta [C{{O}_{2}}]+\Delta [CO]} \right)\cdot \left( \frac{M{{W}_{P}}}{M{{W}_{C}}} \right)\cdot {{w}_{C}}\cdot {{\rho }_{f}}\;,\end{eqnarray} \tag{ 1 }$$

where ${{E}_{P}}$ is the emission factor of pollutant P in unit of mg kg⁻¹ of fuel burned. Δ[P], Δ[CO₂ ] and Δ[CO] represent the increase in concentration of pollutant P, CO₂ and CO between the tunnel inlet and outlet, respectively. MW_p and MW_c are molecular weight of pollutant P and carbon, respectively. w_c  = 0.85 is the mass fraction of carbon of gasoline fuel and ρ_f is the gasoline density (here ρ_f  = 740 g L⁻¹). Since in the vehicle exhausts contribution of hydrocarbons to the total carbon concentrations is known to be negligible compared to that of CO₂ (Kirchstetter et al 1998), hydrocarbons have been ignored in equation (1). As measured CO₂ and CO concentrations were 2 h averages for two days, NO_x and NH₃ concentrations recorded during the same time intervals at the tunnel inlet and outlet were also averaged to calculate Δ[P]. As shown in figure 2(d), percentages of light-duty (LD) vehicles (cars, taxi, MTs, LDTs and LDBs) and heavy-duty (HD) vehicles (MDTs, HDTs, MDBs and HDBs) were relatively stable from 10 am to 10 pm, so data from 10 am to 10 pm were used to calculate the emission factors of NO_x and NH₃.

Given that diesel trucks are found to contribute very little to ammonia (Burgard et al 2006) and that gasoline powered vehicles shared an average of 88.2% in the total traffic densities in the Zhujiang tunnel, gasoline powered vehicles would be the main source of ammonia in the tunnel. Diesel trucks did contribute to CO₂ and might have different fuel efficiencies with gasoline powered vehicles. The fraction of fuel burned by light-duty gasoline vehicles is needed when calculating the ammonia emission factor. Using the fuel efficiencies of different kinds of vehicles reported by He et al (2005) and Huo et al (2012) and the fleet distribution in the tunnel, we obtain a rough estimate of 74.2% of fuel burned by light-duty gasoline vehicles. The emission factor of NH₃ is thus determined to be 2.92 ± 0.18 g L⁻¹ (mean ± 95% C.I.) when all ammonia emission is attributed to light-duty gasoline vehicles. Reported on-road or dynamometer-based NH₃ emission factors for motor vehicles in other countries ranged 88.8–503.2 mg L⁻¹ (Moeckli et al 1996, Fraser and Cass 1998, Kean et al 2000, Baum et al 2001, Durbin et al 2002, 2004, Huai et al 2003, Burgard et al 2006), less than 1/5 of that determined in this study. Recently, Carslaw and Rhys-Tyler (2013) reported an NH₃ emission factor 0.48 g L⁻¹ for Euro IV passenger cars in London, and Sun et al (2014) observed an average on-road NH₃ emission factor 0.36 ± 0.04 g L⁻¹ in California. These factors were quite near that reported by Fraser and Cass (1998) and Kean et al (2000).

Ammonia emission factor based on mg km⁻¹ can be estimated using the fuel efficiency. Wagner et al (2009) estimated the 2006 average fuel consumption of cars in China to be 7.95 L/100 km and projected a 2009 new passenger car average fuel consumption of 7.87 L/100 km. Using 7.87 L/100 km as the average fuel efficiency, we obtained ammonia emission factor of 230 ± 14 mg km⁻¹ (mean ± 95% C.I.) for the light-duty gasoline vehicles. A comparison of ammonia emission factor with those from other tunnel studies is given in table 2. Also ammonia emission factor in mg km⁻¹ measured in the Zhujiang tunnel was the highest among the studies (Moeckli et al 1996, Fraser and Cass 1998, Kean et al 2000, Emmenegger et al 2004).

Since NH₃ is present in human breath in levels ranging 100–2000 ppb (Španěl et al 1998, Kearney et al 2002), how much NH₃ could be attributed to passengers inside the vehicles? Assuming inhalation rate of 10–15 L min⁻¹ for an adult, and taking the NH₃ level in human breath as its upper limit of 2000 ppb, the emission of NH₃ would be 13.7–20.3 μg min⁻¹ for an adult, quite near that of 18.7 μg min⁻¹ as reported by Ament et al (1999). Even with the emission rate of 20.3 μg min⁻¹, emission factor of NH₃ from human breath would be 0.16 mg km⁻¹ assuming an average of five passengers per vehicle and an average vehicle speed of 45 km h⁻¹ in the tunnel. This factor was negligible compared to the 230 mg km⁻¹ from vehicles. Moreover, vehicles passing the tunnels had their windows closed, with a quite limited portion of NH₃ from human breath entering the tunnel through indoor/outdoor exchange. The higher gasoline sulfur levels in China may impact the NH₃ emission. As observed by Durbin et al (2004), NH₃ emission rates were approximately five times higher over the more aggressive US06 test cycle compared to the federal test procedure (FTP) cycle, and sulfur did not affect NH₃ emissions over the FTP, but higher NH₃ emissions were found for increasing fuel sulfur levels over the US06. Malfunctioning three-way-catalysts might be a major factor causing higher ammonia emissions. Fraser and Cass (1998) reported ammonia emission factors ranging from 67.0 to 166.6 mg km⁻¹ for light-duty vehicles equipped with malfunctioning three-way-catalysts. Nevertheless, it is a question that needs further investigation whether the much higher NH₃ emission factor in this study was resulted from fuel quality, performance of catalytic converters or driving conditions.

NO_x (as NO) emission factor was determined to be 7.17 ± 0.60 g L⁻¹ (0.56 ± 0.05 g km⁻¹) (mean ± 95% C.I.) when averaged over the entire fleet in the Zhujiang tunnel. As shown in table 2, NO_x emission factor in the Zhujiang tunnel is within the range of other studies. Che et al (2009) estimated a NO_x emission factor of 1.40 g km⁻¹ for light-duty gasoline vehicles in 2006 in the PRD region using MOBILE and PART5 model, and Liao et al (2011), using COPERT IV model, estimated a NO_x emission factor of 0.51 g km⁻¹ for light-duty gasoline vehicles in Guangzhou in 2008. Our NO_x emission factor was quite near that by Liao et al (2012) but less than half of that by Che et al (2009).

The most comprehensive NH₃ emission inventory for on-road mobile source in the PRD region was recently compiled by Zheng et al (2012). This inventory, however, was based on the emission factors taken from the Emission Inventory Improvement Program (EIIP) by Reo et al (2004), which were 63.2, 4.2, 28.0, 16.8, and 7.0 mg km⁻¹ for light-duty gasoline, light-duty diesel, heavy-duty gasoline, heavy-duty diesel vehicles, and motorcycles, respectively. They obtained NH₃ emission of 4.9 kt yr⁻¹, or 2.5% of the total NH₃ emissions, from on-road mobile source in the PRD region in 2006. If we just replace the emission factor for light-duty gasoline vehicles with 230 mg km⁻¹ determined in this study, NH₃ emission from on-road mobile sources by Zheng et al (2012) would then reach 16.8 kt yr⁻¹ and account for 8.1% of the total NH₃ emission in the region, comparable to 10% obtained by Yao et al (2013) near a highway with the highest traffic density in Canada. As all newly manufactured light-duty vehicles in China are equipped with TWC for exhaust control, and meanwhile in recent decades vehicle population in the PRD region is increasing rapidly at a rate of over 10% per year (Che et al 2011), NH₃ emission from vehicle exhausts would become increasingly important. Applying this emission factor to the vehicle population in 2012 in the PRD region, for NH₃ emission from on-road mobile source we would get an estimate of 44.2 kt yr⁻¹, which is about 2.6 times that in 2006. If the total agricultural NH₃ emissions in 2012 were the same as that in 2006 as estimated by Zheng et al (2012), NH₃ emission from vehicle exhausts would occupy a share as high as 18.8% in the total NH₃ emission. In urban areas where on-road vehicle density is relatively higher and there is much less agricultural activity, NH₃ emission from on-road vehicles would be more distinctive and thus greatly influence the near-surface atmospheric chemistry. Further studies are needed to make clear the reason for the much higher emission factors determined from our tunnel tests in Guangzhou.